[linux-lvm] pvmove ==> segfault: pvs --all -o +vg_fmt
Martin Papik
mp6058 at gmail.com
Wed Jul 24 03:33:16 UTC 2013
No reply yet so I thought I'd modify the subject to catch someone's eye.
Martin
On 07/23/2013 05:45 AM, Martin Papik wrote:
>
> Thank you. Is that patch going to be added to the main source code base?
>
> I think I found an error (segfault), see below.
>
> ==> I have a few more improvement suggestions.
>
> * IMHO vgs/lvs should include a column called seg_size_pe, similar to
> seg_start and seg_start_pe
> * IMHO vgs/lvs/pvs should have an command line parameter, or option
> modifier allowing to not display empty columns, i.e. the following
> command might skip displaying origin, snap%, move, log, copy%,
> convert, because they have no values. That would allow me to show
> copy_percent column only when copy is in progress. Would be nice.
>
> lvs --all -o +vg_extent_size
> LV VG Attr LSize Origin Snap% Move Log Copy% Convert Ext
> boot volume -wi-ao 1.00g 4.00m
> copy volume -wi-ao 2.00t 4.00m
> emulate volume -wi-ao 50.00g 4.00m
> home volume -wi-ao 150.00g
> 4.00m
> p2p volume -wi-a- 300.00g
> 4.00m
> root volume -wi-ao 4.00g 4.00m
> swap volume -wi-ao 8.00g 4.00m
> tmp volume -wi-ao 2.00g 4.00m
> usr volume -wi-ao 20.00g 4.00m
> var volume -wi-ao 20.00g 4.00m
>
> Possible error:
>
> root at localhost# lvs --rows --align
> Does not align, should it?
>
> *==> SEGFAULT INFO*
>
> root at localhost# pvs --all -o +vg_fmt
> Segmentation fault (core dumped)
>
> However the following seems to work.....
>
> root at localhost# pvs --all -o +vg_name
> PV VG Fmt Attr PSize PFree VG
> /dev/dm-0 -- 0 0
> --- some entries removed for brevity ----
> /dev/md127 volume lvm2 a- 5.46t 2.77t volume
> --- some entries removed for brevity ----
> /dev/sde1 -- 0 0
>
> Additional information that might help.
>
> root at localhost# pvs --version
> LVM version: 2.02.66(2) (2010-05-20)
> Library version: 1.02.48 (2010-05-20)
> Driver version: 4.22.0
> root at localhost# cat /etc/lsb-release
> DISTRIB_ID=Ubuntu
> DISTRIB_RELEASE=12.04
> DISTRIB_CODENAME=precise
> DISTRIB_DESCRIPTION="Ubuntu 12.04.2 LTS"
>
> root at localhost#valgrind pvs --all -o +vg_fmt
> ==14584== Memcheck, a memory error detector
> ==14584== Copyright (C) 2002-2011, and GNU GPL'd, by Julian Seward et al.
> ==14584== Using Valgrind-3.7.0 and LibVEX; rerun with -h for copyright
> info
> ==14584== Command: pvs --all -o +vg_fmt
> ==14584==
> ==14584== Invalid read of size 8
> ==14584== at 0x548ED77: dm_report_field_string (in
> /lib/libdevmapper.so.1.02.1)
> ==14584== by 0x549006F: dm_report_object (in
> /lib/libdevmapper.so.1.02.1)
> ==14584== by 0x46B138: report_object (in /sbin/lvm)
> ==14584== by 0x423618: ??? (in /sbin/lvm)
> ==14584== by 0x4254B2: process_each_pv (in /sbin/lvm)
> ==14584== by 0x4230D3: ??? (in /sbin/lvm)
> ==14584== by 0x41AE60: lvm_run_command (in /sbin/lvm)
> ==14584== by 0x41B307: lvm2_main (in /sbin/lvm)
> ==14584== by 0x58CE76C: (below main) (libc-start.c:226)
> ==14584== Address 0x28 is not stack'd, malloc'd or (recently) free'd
> ==14584==
> ==14584==
> ==14584== Process terminating with default action of signal 11 (SIGSEGV)
> ==14584== Access not within mapped region at address 0x28
> ==14584== at 0x548ED77: dm_report_field_string (in
> /lib/libdevmapper.so.1.02.1)
> ==14584== by 0x549006F: dm_report_object (in
> /lib/libdevmapper.so.1.02.1)
> ==14584== by 0x46B138: report_object (in /sbin/lvm)
> ==14584== by 0x423618: ??? (in /sbin/lvm)
> ==14584== by 0x4254B2: process_each_pv (in /sbin/lvm)
> ==14584== by 0x4230D3: ??? (in /sbin/lvm)
> ==14584== by 0x41AE60: lvm_run_command (in /sbin/lvm)
> ==14584== by 0x41B307: lvm2_main (in /sbin/lvm)
> ==14584== by 0x58CE76C: (below main) (libc-start.c:226)
> ==14584== If you believe this happened as a result of a stack
> ==14584== overflow in your program's main thread (unlikely but
> ==14584== possible), you can try to increase the size of the
> ==14584== main thread stack using the --main-stacksize= flag.
> ==14584== The main thread stack size used in this run was 8388608.
> ==14584==
>
> root at localhost# ltrace pvs --all -o +vg_fmt 2>&1 | tail
> __ctype_toupper_loc() = 0x7f2f0e83f768
> __ctype_toupper_loc() = 0x7f2f0e83f768
> __ctype_toupper_loc() = 0x7f2f0e83f768
> __sprintf_chk(0x1e86e88, 1, -1, 0x48daa5, 0x48dab8) = 2
> dm_pool_strdup(0x1e88120, 0x1e86e88, 0x1e86e8a, 0, 0x48daa8) = 0x1e8e538
> dm_pool_alloc(0x1e88120, 8, 2, 0x7f2f0d892030, 0x48daa8) = 0x1e8e540
> dm_report_field_set_value(0x1e8e508, 0x1e8e538, 0x1e8e540,
> 0x7f2f0d892030, 0x48daa8) = 0x1e8e540
> dm_report_field_string(0x1e88080, 0x1e8e550, 40, 0x6b1fe0, 0x1e6cfe0
> <unfinished ...>
> --- SIGSEGV (Segmentation fault) ---
> +++ killed by SIGSEGV +++
> root at localhost#
>
> root at localhost# gdb --args pvs --all -o +vg_fmt
> GNU gdb (Ubuntu/Linaro 7.4-2012.04-0ubuntu2.1) 7.4-2012.04
> Copyright (C) 2012 Free Software Foundation, Inc.
> License GPLv3+: GNU GPL version 3 or later
> <http://gnu.org/licenses/gpl.html>
> This is free software: you are free to change and redistribute it.
> There is NO WARRANTY, to the extent permitted by law. Type "show copying"
> and "show warranty" for details.
> This GDB was configured as "x86_64-linux-gnu".
> For bug reporting instructions, please see:
> <http://bugs.launchpad.net/gdb-linaro/>...
> Reading symbols from /sbin/pvs...(no debugging symbols found)...done.
> (gdb) run
> Starting program: /sbin/pvs --all -o +vg_fmt
> [Thread debugging using libthread_db enabled]
> Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
> File descriptor 3 (socket:[46992174]) leaked on pvs invocation. Parent
> PID 14623: gdb
> File descriptor 4 (socket:[46992175]) leaked on pvs invocation. Parent
> PID 14623: gdb
> File descriptor 5 (pipe:[46992176]) leaked on pvs invocation. Parent
> PID 14623: gdb
> File descriptor 6 (pipe:[46992176]) leaked on pvs invocation. Parent
> PID 14623: gdb
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x00007ffff757dd77 in dm_report_field_string () from
> /lib/libdevmapper.so.1.02.1
> (gdb) disassemble dm_report_field_string
> Dump of assembler code for function dm_report_field_string:
> 0x00007ffff757dd70 <+0>: push %rbx
> 0x00007ffff757dd71 <+1>: mov %rsi,%rbx
> 0x00007ffff757dd74 <+4>: mov (%rdi),%rdi
> => 0x00007ffff757dd77 <+7>: mov (%rdx),%rsi
> 0x00007ffff757dd7a <+10>: callq 0x7ffff7570a30 <dm_pool_strdup at plt>
> 0x00007ffff757dd7f <+15>: test %rax,%rax
> (gdb) info registers
> rax 0x6dd550 7198032
> rbx 0x6dd550 7198032
> rcx 0x6b1fe0 7020512
> rdx 0x28 40
> rsi 0x6dd550 7198032
> rdi 0x6d7120 7172384
> rbp 0x6d7080 0x6d7080
>
>
>
> On 07/20/2013 02:20 AM, Alasdair G Kergon wrote:
>> It uses lists of ranges separated by colons.
>>
>> :x-y
>>
>> If x is missing, 0 is inserted; if y is missing, the last extent on the device
>> is inserted.
>>
>> Man pages can always be improved with more examples!
>>
>> For +length, try the patch below.
>>
>> Alasdair
>>
>>
>> --- a/tools/toollib.c
>> +++ b/tools/toollib.c
>> @@ -1001,7 +1001,7 @@ static int _parse_pes(struct dm_pool *mem, char *c, struct dm_list *pe_ranges,
>> const char *pvname, uint32_t size)
>> {
>> char *endptr;
>> - uint32_t start, end;
>> + uint32_t start, end, len;
>>
>> /* Default to whole PV */
>> if (!c) {
>> @@ -1041,7 +1041,16 @@ static int _parse_pes(struct dm_pool *mem, char *c, struct dm_list *pe_ranges,
>> goto error;
>> c = endptr;
>> }
>> + } else if (*c == '+') { /* Length? */
>> + c++;
>> + if (isdigit(*c)) {
>> + if (!xstrtouint32(c, &endptr, 10, &len))
>> + goto error;
>> + c = endptr;
>> + end = start + (len ? (len - 1) : 0);
>> + }
>> }
>> +
>> if (*c && *c != ':')
>> goto error;
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-lvm/attachments/20130724/e3b4e49c/attachment.htm>
More information about the linux-lvm
mailing list